Paper 2602.19672
SkillOrchestra: Learning to Route Agents via Skill Transfer
- Published
- Feb 2026
- Research lab
- Independent
- Citations
- 16
- GitHub
- 71 stars
01 In brief
Summary
SkillOrchestra is a framework for skill-aware orchestration in compound AI systems, addressing limitations of existing routing approaches: input-level routers make coarse decisions, and RL-trained orchestrators are expensive and prone to routing collapse.
Instead of learning a routing policy end-to-end, SkillOrchestra learns a reusable Skill Handbook from execution traces, containing mode-level insights, fine-grained skills, and agent profiles with competence and cost estimates.
At deployment, the orchestrator selects an operational mode, identifies active skills, and chooses agents based on a performance-cost trade-off.
Experiments across ten benchmarks show SkillOrchestra outperforms state-of-the-art RL-based orchestrators by up to 22.5% accuracy, with 700x and 300x learning cost reduction compared to Router-R1 and ToolOrchestra, respectively.
It also mitigates routing collapse, achieving balanced model utilization, and transfers across orchestrator backbones without retraining.
The framework includes a Pareto-optimal handbook selection step to match skill granularity to orchestrator capacity.
Ablations confirm that skill discovery, refinement, and selection are crucial for performance and cost efficiency.
02 From the paper
Abstract
Compound AI systems promise capabilities beyond those of individual models, yet their success depends critically on effective orchestration. Existing routing approaches face two limitations: (1) input-level routers make coarse query-level decisions that ignore evolving task requirements; (2) RL-trained orchestrators are expensive to adapt and often suffer from routing collapse, repeatedly invoking one strong but costly option in multi-turn scenarios. We introduce SkillOrchestra, a framework for skill-aware orchestration. Instead of directly learning a routing policy end-to-end, SkillOrchestra learns fine-grained skills from execution experience and models agent-specific competence and cost under those skills. At deployment, the orchestrator infers the skill demands of the current interaction and selects agents that best satisfy them under an explicit performance-cost trade-off. Extensive experiments across ten benchmarks demonstrate that SkillOrchestra outperforms SoTA RL-based orchestrators by up to 22.5% with 700x and 300x learning cost reduction compared to Router-R1 and ToolOrchestra, respectively. These results show that explicit skill modeling enables scalable, interpretable, and sample-efficient orchestration, offering a principled alternative to data-intensive RL-based approaches. The code is available at: https://github.com/jiayuww/SkillOrchestra.